基于进化算法的带约束混合动力系统多目标优化

Constrained Parallel Hybrid System Multi-objective Optimization Based on Evolutionary Algorithm

为获得无需将多目标优化问题转化为单目标优化问题的混合动力系统多目标优化方法,分析了并联混合汽车总成模型,建立了带约束混合动力系统多目标优化数学模型,并给出了优化目标、待优化参数及约束条件。设计了基于NSGA-Ⅱ的混合动力系统多目标优化算法,该算法基于Pareto支配性原理判定所得方案的优劣,不需要指定各个目标的权系数。仿真优化结果表明:优化后的系统百公里油耗平均下降了0.25%,污染物排放平均下降了2.75%,蓄电池充电效率分布由[0.8,0.9]变为[0.85,0.9],放电效率分布由[0.82,1.0]变为[0.95,1.0],作者提出的方法可以优化混合动力系统的性能。

In order to obtain a method to avoid transforming multi-objective functions into a single objective evaluation function for hybrid system multi-objective optimization problem,the parallel hybrid electric vehicle model was analyzed,the multi-objective optimization mathematical model of constrained hybrid system was established,and the optimization objectives,parameters and constraints were given.A multi-objective evolutionary algorithm for constrained parallel hybrid system optimization based on NSGA-Ⅱ（cPHS-NSGA） was proposed,which adopted the Pareto dominated principle to determine solutions without specifying weight coefficient for each objective.The simulation optimization results showed that compared with the old system,the fuel consumption per 100 km dropped by an average of 0.25% and the emissions fell by an average of 2.75%.Battery charging efficiency distribution changed from [0.8,0.9] to [0.85,0.9] and the range of discharge efficiency changed from [0.82,1.0] to [0.95,1.0].The cPHS-NSGA was capable to improve the performance of parallel hybrid system.